Does Every Hospital Admission Deserve a Root Cause Analysis?

In a system of value-based care, a hospital admission can be considered a failure to keep a person healthy. Is that hyperbolic? Many hospitalizations are “no fault”: the onset of disease with factors outside of our control or an unpreventable injury. But the universe of preventable admissions is large. Does our current documentation of a hospitalization systematically probe, poke, and provoke about causation? Does it get to those difficult social determinants of health that can be at the root of an admission?

The practice of root cause analysis (RCA) offers such an approach. And it gives health care providers a rich tool with which to involve our patients and families in their care.

The RCA process is familiar in hospital quality and risk management, where it is used to analyze medical errors, adverse outcomes, and process failures. The goal of an RCA is to identify causal factors and implement solutions that will prevent the incident under analysis from happening again. Root causes of medical errors are classified into domains.

Today, a clinical hospital note (e.g., SOAP note format: Subjective, Objective, Assessment, Plan) ends with a list of diagnoses that are easily coded, billed out, and available for data analysis. This format, with variations, has not changed in decades of medical training and practice.

An RCA approach offers a different set of “diagnoses” — domains, in fact — which encompass social determinants of health and systemic determinants of care delivery operation that could be used to address populations and policy. The RCA outcome is a toolbox of actions that will help prevent the next hospital admission.

Naturally, some root causes for hospitalizations will be simple, falling under the domains of “Disease Occurrence, No Fault” (e.g., diverticular abscess, brain metastasis) or “Trauma, No Fault” (e.g., fractures sustained in an unprovoked assault). In these cases, there is little room for root cause recommendation aside from standard medical/surgical care.

But trauma can also provide quick examples where actionable root cause domains, if we look for them, can often be identified:

A case review is illustrative: A 45-year-old with a history of tobacco use, BMI = 40, and a lower extremity DVT 3 years prior is admitted for hospitalization with acute-on-subacute pulmonary emboli, right heart failure, and hypoxia. The Discharge Problem List mentions all of these, along with tobacco dependence.

A review of this hospitalization using the tools of Root Cause Analysis provides a retrospective time line and a patient-centered perspective, one with a set of identifiable root cause domains, solutions, and recommendations. It also answers the question, “Could this hospital admission have been avoided?” The answer is “Yes”.

Root Cause Analysis:

1.  Define the problem

• Pulmonary emboli, acute-on-chronic, now with right heart failure and hypoxia

2.  Collect data

Primary care records, specialty care records, Emergency Department records, interview(s) with patient, family, and providers

3.  Identify possible causal factors

(sequence of events, conditions that allowed problem to occur; other problems that contributed to central problem)

• Primary Cause
  • Patient went off Medicare Part D 9 months ago, could not afford rivaroxaban through time of this admission; did not advocate for assistance
• Secondary Causes
  • Patient did not understand the disease (venous thrombosis), medical management, and critical nature of medication compliance
  • Patient did not appreciate that her primary care health team “was there to help” with the Primary Cause
  • Patient noncompliance with a critical medication not fully appreciated by primary health care team when referenced on multiple outpatient and ED visits
  • Primary recommendation for chronic anticoagulation “buried” in EMR
  • Continued tobacco use
  • Continued BMI > 30; deconditioning

4.  Identify root causes and assign root cause domains

• Insurance coverage lapse leading to medication noncompliance
  • Access to Care – Financial
• Failure of patient education about condition, role of critical medication, role of health team
  • Patient/family Understanding of Medical Condition
  • Patient/family Understanding of Medical Management
  • Care Coordination- Patient Disengagement with Treatment Team
• Failure of health team to identify and follow up with medication noncompliance
  • Care Coordination- Establishment & Implementation of Care Plan
  • Efficacy of the Medical Record
• Electronic medical record did not sensitize critical recommendation (chronic anticoagulation)
  • Efficacy of the Medical Record
• Electronic medical record did not alert to critical medication noncompliance
  • Efficacy of the Medical Record
• Lifestyle contributions to current and ongoing morbidity
  • Modifiable behavior – Substance Use Disorder (Tobacco)
  • Modifiable behavior – Obesity
  • Modifiable behavior – Physical Conditioning/Exercise

5.  Recommend and implement solutions

• Patient assistance and education regarding health insurance coverage security, ongoing
• Primary Care Health Team to perform process improvement to better identify medication noncompliance early and intervene promptly
• Primary Care Health Team review of patient’s care coordination needs
• Patient education about disease and medical management, and ongoing
• Patient education about role of Primary Care Health Team
• Assess if EMR can be modified to highlight critical medications, critical recommendations, and evidence of patient compliance
• Primary Care Health Team should work with patient to motivate and coordinate ongoing tobacco cessation therapy, exercise, and weight loss

The plan of care after an RCA is, therefore, differently documented and directed when compared with a traditional SOAP note, and redounds in new ways to the benefit of the patient. In a population, the domain data gathered would accrue to demonstrate more specific and actionable root determinants of hospitalizations than we measure systematically today.

An important extension is to involve patients and families/caregivers in these analyses! Their input and narratives will be key to the analysis, and they are the primary stakeholders in changing many root causes. Their involvement in such a rational objective process can be, and must be, both educational and motivational.

Implementing the Root Cause Analysis

A hospitalist group or Internal Medicine residency training program would be best situated to pilot application of RCA. Several questions should be tested:

• What skills are needed to get to a successful RCA in this new context? Would a skilled nurse/case manager, physician, or an advanced practice provider each independently arrive at the same data and conclusions? Would a team approach work best as most traditional RCAs are conducted? Which approach is most cost effective?
• How are outcomes measured over time? A controlled trial would be ideal, with primary end points being hospital admissions and perhaps even ED visits.
• How would hospital documentation need to be modified to include the RCA, in such a way that the domains are used as data points for population health analytics and research?
• How are the techniques of the RCA taught? This would be most effective early in education and reinforced throughout postgraduate training.

As an Emergency Medicine physician, I can’t help but look at the RCA concept and consider adapting it to most proximate causation. Providers would seek to quickly identify those condition(s) that could bring about an achievable action plan.

Here in rural Vermont where I practice, our Emergency Department group is working on such an approach. We are starting with the construction in our EMR of documentation elements for “Complexity of Care” encompassing sub-factors of “Care Coordination” and “Counseling” tied to time spent with the patient. These time-consuming elements of care undertaken by an ED provider will need to be justified by value-based impacts on wellness and risk mitigation. Adoption of a most proximate RCA process to an ED visit in coordination with our Patient-Centered Medical Homes then makes sense. We will look to test the questions outlined above.

If successful, we could add a couple more interesting pieces of the puzzle to the population health strategy being steered by our statewide ACO, OneCare Vermont.

In sum, applying the RCA rubric to hospital admissions may better help define and manage the care of individuals and populations, prevent costly utilization, and potentially inform policy.